When someone asks ChatGPT for "a good accountant in Alexandria," the model doesn't search the way Google does. Each AI engine leans on a different local-data pipe — and if your business isn't in the right pipe, you don't exist in that engine's answer. Here's the map.
A common mistake is treating "AI visibility" as one thing. In practice, ChatGPT, Perplexity, Microsoft Copilot, and Google's AI features each resolve a local-business question against a different primary data source. Your website matters to all of them — but for local recommendations specifically, the deciding factor is usually whether you have a complete, consistent profile in the directory that particular engine trusts. The good news: claiming those profiles is cheap, fast, and mostly a one-time job.
The short version
- Each engine has a dominant local-data source: reported pairings are ChatGPT ↔ Foursquare, Perplexity ↔ Yelp, Copilot ↔ Bing Places, Google AI ↔ Google Business Profile.
- Review volume and rating act as an inclusion filter, not just decoration — engines skip thinly reviewed listings.
- NAP consistency (identical name, address, phone everywhere) is what lets an engine trust that all your profiles are one business.
- Claim all four core profiles once, keep them identical, and you've covered every major engine's data pipe.
Why AI engines need directories at all
Large language models don't hold a live index of every plumber, dentist, and marketing agency. When an assistant answers a local-recommendation question, it retrieves candidate businesses from structured places data — licensed databases and public directories with names, categories, locations, hours, and reviews — then writes its recommendation from what it retrieved. That makes the retrieval layer, not the model, the gatekeeper. A brilliant website attached to an unclaimed, half-empty directory presence is invisible at exactly the moment the engine is choosing whom to name.
This is entity SEO applied locally: the engine needs to resolve "your business" into a single, trustworthy record before it will put your name in an answer. Every consistent profile is another vote that the record is real.
The engine-by-engine map
The pairings below are drawn from industry reporting and platform documentation. Exact market-share figures floating around vendor blogs are unverified, so we present the qualitative pattern — which is consistent across sources — rather than precise percentages.
| Engine | Reported primary local-data source | What to claim | Confidence |
|---|---|---|---|
| ChatGPT | Foursquare's places database (reported by multiple sources as its main local pipe), plus Bing's web index for site content | Foursquare for Business listing; keep your site indexable by Bing | Medium — relationship widely reported, exact share unverified |
| Perplexity | Yelp listings and reviews, plus its own live web retrieval | Yelp business page with active reviews | Medium — consistent across industry reporting |
| Copilot / Bing | Bing Places directly, cross-checked against NAP data on the open web | Bing Places (imports from Google Business Profile) + Bing Webmaster Tools | High — Microsoft's own ecosystem |
| Google AI Overviews / Gemini | Google Business Profile and the Knowledge Graph it feeds | Google Business Profile, fully completed, with reviews | High — Google's own ecosystem |
ChatGPT: the Foursquare surprise
The least obvious entry on that map is the most important one. Foursquare stopped being a check-in app years ago and quietly became a business-to-business places-data company; its database of roughly 100 million points of interest is reported to supply location data to Apple Maps, Uber, Samsung — and OpenAI. Multiple industry sources describe Foursquare as ChatGPT's primary source for local-business results. The exact share is unverified, but the relationship itself is corroborated widely enough that an unclaimed Foursquare listing is an unforced error: it's free, takes minutes, and may be the single pipe ChatGPT checks first.
ChatGPT also reads the open web through Bing's index when it browses, which is why the same page structure that wins citations elsewhere — answer-first sections, extractable passages, real facts — still matters. Our eight-step playbook covers that side of the work.
Perplexity: reviews are the currency
Perplexity's local answers lean heavily on Yelp — both the listing data and the review content, which the engine quotes and paraphrases when it explains why it recommends a business. That changes the job: a bare Yelp profile with two reviews doesn't just look thin to humans, it gives Perplexity nothing to say about you. A steady flow of genuine, detailed reviews is content marketing for the AI era, written by your customers.
Copilot: Bing Places and the consistency check
Microsoft Copilot draws on Bing Places, and Bing's systems are described as cross-checking a listing's name, address, and phone number against mentions across the open web as a trust signal. That's the practical argument for NAP consistency: when your phone number is formatted three different ways across five directories, the engine can't be sure it's one business — and uncertain entities don't get recommended. Bing Places can import your Google Business Profile, which makes it one of the fastest claims on the list.
Google: the profile is the product
For AI Overviews and Gemini, the local layer is Google Business Profile feeding the Knowledge Graph. This is the one most local businesses have already claimed — but "claimed" and "complete" are different things. Category, services, service areas, hours, photos, and above all reviews determine whether the AI layer treats your listing as a confident answer or skips it. If you want the box itself, we've written a full field guide to appearing in AI Overviews.
Reviews: the filter nobody mentions
Across every engine, review volume and rating behave like an inclusion threshold rather than a tiebreaker. Independent research backs the intuition that AI engines filter harder than local search does: the SOCi 2026 Local Visibility Index found only 1.2% of 350,000+ business locations were recommended by ChatGPT, versus 35.9% appearing in Google's local 3-pack. The engines are choosing a shortlist, not ranking everyone — and sparse, low-rated listings don't make shortlists.
What to do this week
The whole map collapses into one afternoon of work, done once and maintained lightly:
1. Write your canonical NAP block — exact business name, locality, phone, email, website — and never deviate from it. 2. Claim the four core profiles: Google Business Profile, Foursquare for Business, Bing Places, and Yelp, all with identical details. 3. Start collecting reviews on Google and Yelp immediately; volume compounds. 4. Link the profiles from your website's Organization schema (the sameAs field) so engines can verify that the site and the profiles are one entity. Then measure: ask each engine your own customers' questions monthly and log whether you're named — the trend is your real AI-visibility metric.
Common questions
How does ChatGPT find local businesses?
ChatGPT retrieves local-business candidates from licensed places data — Foursquare's database is reported by multiple industry sources to be its primary local source — and supplements that with web content from Bing's index when browsing. To be findable, claim and complete a Foursquare for Business listing and keep your site crawlable by Bing.
Why does my business show up on Google but not in ChatGPT or Perplexity?
Because they use different data pipes. Google's AI features draw on your Google Business Profile, while ChatGPT is reported to lean on Foursquare and Perplexity on Yelp. A business that has only claimed its Google profile is invisible in the directories the other engines check first.
Do online reviews affect AI recommendations?
Yes — review volume and rating act as an inclusion filter. AI engines recommend a short list, not a full ranking, and listings with few or poor reviews tend to be skipped entirely. Independent research found ChatGPT recommends only about 1.2% of business locations when asked, a far stricter filter than Google's local 3-pack.
What is NAP consistency and why does it matter for AI?
NAP is your business Name, Address, and Phone number. AI engines cross-check these details across directories and the open web to confirm that all the profiles describe one real business. Inconsistent details fragment your entity and lower the engine's confidence, which costs you recommendations.
Which business profiles should a local business claim for AI visibility?
Four cover the major engines: Google Business Profile (Google AI Overviews and Gemini), Foursquare for Business (reported ChatGPT source), Bing Places (Microsoft Copilot), and Yelp (Perplexity). Use identical business details on all four and link them from your website's Organization schema.